From 969d7682053bb73cc78503556d63dd10039eb59f Mon Sep 17 00:00:00 2001
From: "Meredith L. Patterson" <mlp@thesmartpolitenerd.com>
Date: Sat, 3 Oct 2015 17:54:11 +0200
Subject: [PATCH] Don't cache dummy results for primitive parsers.
---
src/backends/packrat.c | 30 +++++++++++++++++-------------
src/internal.h | 7 ++++---
src/parsers/action.c | 1 +
src/parsers/and.c | 1 +
src/parsers/attr_bool.c | 1 +
src/parsers/bind.c | 1 +
src/parsers/bits.c | 1 +
src/parsers/butnot.c | 1 +
src/parsers/ch.c | 1 +
src/parsers/charset.c | 1 +
src/parsers/choice.c | 1 +
src/parsers/difference.c | 1 +
src/parsers/end.c | 1 +
src/parsers/endianness.c | 1 +
src/parsers/epsilon.c | 1 +
src/parsers/ignore.c | 1 +
src/parsers/ignoreseq.c | 1 +
src/parsers/indirect.c | 1 +
src/parsers/int_range.c | 1 +
src/parsers/many.c | 1 +
src/parsers/not.c | 1 +
src/parsers/nothing.c | 1 +
src/parsers/optional.c | 1 +
src/parsers/permutation.c | 1 +
src/parsers/sequence.c | 1 +
src/parsers/token.c | 1 +
src/parsers/unimplemented.c | 1 +
src/parsers/value.c | 2 ++
src/parsers/whitespace.c | 1 +
src/parsers/xor.c | 1 +
30 files changed, 50 insertions(+), 16 deletions(-)
diff --git a/src/backends/packrat.c b/src/backends/packrat.c
index e6f86f29..9977cc3c 100644
--- a/src/backends/packrat.c
+++ b/src/backends/packrat.c
@@ -3,7 +3,7 @@
#include "../internal.h"
#include "../parsers/parser_internal.h"
-// short-hand for creating cache values (regular case)
+// short-hand for creating lowlevel parse cache values (parse result case)
static
HParserCacheValue * cached_result(HParseState *state, HParseResult *result) {
HParserCacheValue *ret = a_new(HParserCacheValue, 1);
@@ -13,7 +13,7 @@ HParserCacheValue * cached_result(HParseState *state, HParseResult *result) {
return ret;
}
-// short-hand for caching parse results (left recursion case)
+// short-hand for creating lowlevel parse cache values (left recursion case)
static
HParserCacheValue *cached_lr(HParseState *state, HLeftRec *lr) {
HParserCacheValue *ret = a_new(HParserCacheValue, 1);
@@ -186,18 +186,22 @@ HParseResult* h_do_parse(const HParser* parser, HParseState *state) {
if (!m) {
// It doesn't exist, so create a dummy result to cache
HLeftRec *base = a_new(HLeftRec, 1);
- base->seed = NULL; base->rule = parser; base->head = NULL;
- h_slist_push(state->lr_stack, base);
- // cache it
- h_hashtable_put(state->cache, key, cached_lr(state, base));
- // parse the input
+ if (parser->vtable->higher) {
+ base->seed = NULL; base->rule = parser; base->head = NULL;
+ h_slist_push(state->lr_stack, base);
+ // cache it
+ h_hashtable_put(state->cache, key, cached_lr(state, base));
+ // parse the input
+ }
HParseResult *tmp_res = perform_lowlevel_parse(state, parser);
- // the base variable has passed equality tests with the cache
- h_slist_pop(state->lr_stack);
- // update the cached value to our new position
- HParserCacheValue *cached = h_hashtable_get(state->cache, key);
- assert(cached != NULL);
- cached->input_stream = state->input_stream;
+ if (parser->vtable->higher) {
+ // the base variable has passed equality tests with the cache
+ h_slist_pop(state->lr_stack);
+ // update the cached value to our new position
+ HParserCacheValue *cached = h_hashtable_get(state->cache, key);
+ assert(cached != NULL);
+ cached->input_stream = state->input_stream;
+ }
// setupLR, used below, mutates the LR to have a head if appropriate, so we check to see if we have one
if (NULL == base->head) {
h_hashtable_put(state->cache, key, cached_result(state, tmp_res));
diff --git a/src/internal.h b/src/internal.h
index b35b7d5a..776f6368 100644
--- a/src/internal.h
+++ b/src/internal.h
@@ -278,9 +278,9 @@ typedef struct HRecursionHead_ {
/* A left recursion.
*
* Members:
- * seed -
- * rule -
- * head -
+ * seed - the HResult yielded by rule
+ * rule - the HParser that produces seed
+ * head - the
*/
typedef struct HLeftRec_ {
HParseResult *seed;
@@ -419,6 +419,7 @@ struct HParserVtable_ {
bool (*isValidCF)(void *env);
bool (*compile_to_rvm)(HRVMProg *prog, void* env); // FIXME: forgot what the bool return value was supposed to mean.
void (*desugar)(HAllocator *mm__, HCFStack *stk__, void *env);
+ bool higher; // false if primitive
};
bool h_false(void*);
diff --git a/src/parsers/action.c b/src/parsers/action.c
index 04eb7a4c..a3243334 100644
--- a/src/parsers/action.c
+++ b/src/parsers/action.c
@@ -81,6 +81,7 @@ static const HParserVtable action_vt = {
.isValidCF = action_isValidCF,
.desugar = desugar_action,
.compile_to_rvm = action_ctrvm,
+ .higher = true,
};
HParser* h_action(const HParser* p, const HAction a, void* user_data) {
diff --git a/src/parsers/and.c b/src/parsers/and.c
index c5c9836d..e07bc9fc 100644
--- a/src/parsers/and.c
+++ b/src/parsers/and.c
@@ -17,6 +17,7 @@ static const HParserVtable and_vt = {
revision. --mlp, 18/12/12 */
.isValidCF = h_false, /* despite TODO above, this remains false. */
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
diff --git a/src/parsers/attr_bool.c b/src/parsers/attr_bool.c
index e8359ab0..f7667740 100644
--- a/src/parsers/attr_bool.c
+++ b/src/parsers/attr_bool.c
@@ -79,6 +79,7 @@ static const HParserVtable attr_bool_vt = {
.isValidCF = ab_isValidCF,
.desugar = desugar_ab,
.compile_to_rvm = ab_ctrvm,
+ .higher = true,
};
diff --git a/src/parsers/bind.c b/src/parsers/bind.c
index f024a82f..7fa821dc 100644
--- a/src/parsers/bind.c
+++ b/src/parsers/bind.c
@@ -60,6 +60,7 @@ static const HParserVtable bind_vt = {
.isValidRegular = h_false,
.isValidCF = h_false,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser *h_bind(const HParser *p, HContinuation k, void *env)
diff --git a/src/parsers/bits.c b/src/parsers/bits.c
index 716524ce..be8f13f1 100644
--- a/src/parsers/bits.c
+++ b/src/parsers/bits.c
@@ -102,6 +102,7 @@ static const HParserVtable bits_vt = {
.isValidCF = h_true,
.desugar = desugar_bits,
.compile_to_rvm = bits_ctrvm,
+ .higher = false,
};
HParser* h_bits(size_t len, bool sign) {
diff --git a/src/parsers/butnot.c b/src/parsers/butnot.c
index f114a1fa..24ece4be 100644
--- a/src/parsers/butnot.c
+++ b/src/parsers/butnot.c
@@ -40,6 +40,7 @@ static const HParserVtable butnot_vt = {
.isValidRegular = h_false,
.isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_butnot(const HParser* p1, const HParser* p2) {
diff --git a/src/parsers/ch.c b/src/parsers/ch.c
index b4386cff..3da1091a 100644
--- a/src/parsers/ch.c
+++ b/src/parsers/ch.c
@@ -47,6 +47,7 @@ static const HParserVtable ch_vt = {
.isValidCF = h_true,
.desugar = desugar_ch,
.compile_to_rvm = ch_ctrvm,
+ .higher = false,
};
HParser* h_ch(const uint8_t c) {
diff --git a/src/parsers/charset.c b/src/parsers/charset.c
index e1a910f8..a4b8c89c 100644
--- a/src/parsers/charset.c
+++ b/src/parsers/charset.c
@@ -76,6 +76,7 @@ static const HParserVtable charset_vt = {
.isValidCF = h_true,
.desugar = desugar_charset,
.compile_to_rvm = cs_ctrvm,
+ .higher = false,
};
HParser* h_ch_range(const uint8_t lower, const uint8_t upper) {
diff --git a/src/parsers/choice.c b/src/parsers/choice.c
index bfc3f904..dd3908ce 100644
--- a/src/parsers/choice.c
+++ b/src/parsers/choice.c
@@ -75,6 +75,7 @@ static const HParserVtable choice_vt = {
.isValidCF = choice_isValidCF,
.desugar = desugar_choice,
.compile_to_rvm = choice_ctrvm,
+ .higher = true,
};
HParser* h_choice(HParser* p, ...) {
diff --git a/src/parsers/difference.c b/src/parsers/difference.c
index 76a2cc44..a24f5acf 100644
--- a/src/parsers/difference.c
+++ b/src/parsers/difference.c
@@ -39,6 +39,7 @@ static HParserVtable difference_vt = {
.isValidRegular = h_false,
.isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_difference(const HParser* p1, const HParser* p2) {
diff --git a/src/parsers/end.c b/src/parsers/end.c
index 30b3ba12..85499d93 100644
--- a/src/parsers/end.c
+++ b/src/parsers/end.c
@@ -25,6 +25,7 @@ static const HParserVtable end_vt = {
.isValidCF = h_true,
.desugar = desugar_end,
.compile_to_rvm = end_ctrvm,
+ .higher = false,
};
HParser* h_end_p() {
diff --git a/src/parsers/endianness.c b/src/parsers/endianness.c
index e3f53ab8..cb3abc3d 100644
--- a/src/parsers/endianness.c
+++ b/src/parsers/endianness.c
@@ -46,6 +46,7 @@ static const HParserVtable endianness_vt = {
.isValidCF = h_false,
.desugar = NULL,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_with_endianness(char endianness, const HParser *p)
diff --git a/src/parsers/epsilon.c b/src/parsers/epsilon.c
index e8ef525f..bb6e8beb 100644
--- a/src/parsers/epsilon.c
+++ b/src/parsers/epsilon.c
@@ -18,6 +18,7 @@ static const HParserVtable epsilon_vt = {
.isValidCF = h_true,
.desugar = desugar_epsilon,
.compile_to_rvm = epsilon_ctrvm,
+ .higher = false,
};
HParser* h_epsilon_p() {
diff --git a/src/parsers/ignore.c b/src/parsers/ignore.c
index af606b0e..c56802ac 100644
--- a/src/parsers/ignore.c
+++ b/src/parsers/ignore.c
@@ -49,6 +49,7 @@ static const HParserVtable ignore_vt = {
.isValidCF = ignore_isValidCF,
.desugar = desugar_ignore,
.compile_to_rvm = ignore_ctrvm,
+ .higher = true,
};
HParser* h_ignore(const HParser* p) {
diff --git a/src/parsers/ignoreseq.c b/src/parsers/ignoreseq.c
index e562136f..07bdc65c 100644
--- a/src/parsers/ignoreseq.c
+++ b/src/parsers/ignoreseq.c
@@ -103,6 +103,7 @@ static const HParserVtable ignoreseq_vt = {
.isValidCF = is_isValidCF,
.desugar = desugar_ignoreseq,
.compile_to_rvm = is_ctrvm,
+ .higher = true,
};
diff --git a/src/parsers/indirect.c b/src/parsers/indirect.c
index 026286d3..b36cb947 100644
--- a/src/parsers/indirect.c
+++ b/src/parsers/indirect.c
@@ -31,6 +31,7 @@ static const HParserVtable indirect_vt = {
.isValidCF = indirect_isValidCF,
.desugar = desugar_indirect,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
void h_bind_indirect__m(HAllocator *mm__, HParser* indirect, const HParser* inner) {
diff --git a/src/parsers/int_range.c b/src/parsers/int_range.c
index 29379930..49b06421 100644
--- a/src/parsers/int_range.c
+++ b/src/parsers/int_range.c
@@ -117,6 +117,7 @@ static const HParserVtable int_range_vt = {
.isValidCF = h_true,
.desugar = desugar_int_range,
.compile_to_rvm = ir_ctrvm,
+ .higher = false,
};
HParser* h_int_range(const HParser *p, const int64_t lower, const int64_t upper) {
diff --git a/src/parsers/many.c b/src/parsers/many.c
index cae2b0ea..6496bbe6 100644
--- a/src/parsers/many.c
+++ b/src/parsers/many.c
@@ -199,6 +199,7 @@ static const HParserVtable many_vt = {
.isValidCF = many_isValidCF,
.desugar = desugar_many,
.compile_to_rvm = many_ctrvm,
+ .higher = true,
};
HParser* h_many(const HParser* p) {
diff --git a/src/parsers/not.c b/src/parsers/not.c
index 6c34bad4..8c2003de 100644
--- a/src/parsers/not.c
+++ b/src/parsers/not.c
@@ -15,6 +15,7 @@ static const HParserVtable not_vt = {
.isValidRegular = h_false, /* see and.c for why */
.isValidCF = h_false,
.compile_to_rvm = h_not_regular, // Is actually regular, but the generation step is currently unable to handle it. TODO: fix this.
+ .higher = true,
};
HParser* h_not(const HParser* p) {
diff --git a/src/parsers/nothing.c b/src/parsers/nothing.c
index 120c1e01..0a60108b 100644
--- a/src/parsers/nothing.c
+++ b/src/parsers/nothing.c
@@ -22,6 +22,7 @@ static const HParserVtable nothing_vt = {
.isValidCF = h_true,
.desugar = desugar_nothing,
.compile_to_rvm = nothing_ctrvm,
+ .higher = false,
};
HParser* h_nothing_p() {
diff --git a/src/parsers/optional.c b/src/parsers/optional.c
index ccee53fa..72660664 100644
--- a/src/parsers/optional.c
+++ b/src/parsers/optional.c
@@ -84,6 +84,7 @@ static const HParserVtable optional_vt = {
.isValidCF = opt_isValidCF,
.desugar = desugar_optional,
.compile_to_rvm = opt_ctrvm,
+ .higher = true,
};
HParser* h_optional(const HParser* p) {
diff --git a/src/parsers/permutation.c b/src/parsers/permutation.c
index 564565af..b1675841 100644
--- a/src/parsers/permutation.c
+++ b/src/parsers/permutation.c
@@ -104,6 +104,7 @@ static const HParserVtable permutation_vt = {
.isValidCF = h_false,
.desugar = NULL,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_permutation(HParser* p, ...) {
diff --git a/src/parsers/sequence.c b/src/parsers/sequence.c
index 93c0cfb9..30de34a4 100644
--- a/src/parsers/sequence.c
+++ b/src/parsers/sequence.c
@@ -93,6 +93,7 @@ static const HParserVtable sequence_vt = {
.isValidCF = sequence_isValidCF,
.desugar = desugar_sequence,
.compile_to_rvm = sequence_ctrvm,
+ .higher = true,
};
HParser* h_sequence(HParser* p, ...) {
diff --git a/src/parsers/token.c b/src/parsers/token.c
index d36ec54b..19029726 100644
--- a/src/parsers/token.c
+++ b/src/parsers/token.c
@@ -73,6 +73,7 @@ const HParserVtable token_vt = {
.isValidCF = h_true,
.desugar = desugar_token,
.compile_to_rvm = token_ctrvm,
+ .higher = false,
};
HParser* h_token(const uint8_t *str, const size_t len) {
diff --git a/src/parsers/unimplemented.c b/src/parsers/unimplemented.c
index e3f30394..e085858b 100644
--- a/src/parsers/unimplemented.c
+++ b/src/parsers/unimplemented.c
@@ -18,6 +18,7 @@ static const HParserVtable unimplemented_vt = {
.isValidCF = h_false,
.desugar = NULL,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
static HParser unimplemented = {
diff --git a/src/parsers/value.c b/src/parsers/value.c
index 531db7cb..7fa863a1 100644
--- a/src/parsers/value.c
+++ b/src/parsers/value.c
@@ -26,6 +26,7 @@ static const HParserVtable put_vt = {
.isValidRegular = h_false,
.isValidCF = h_false,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_put_value(const HParser* p, const char* name) {
@@ -55,6 +56,7 @@ static const HParserVtable get_vt = {
.isValidRegular = h_false,
.isValidCF = h_false,
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_get_value(const char* name) {
diff --git a/src/parsers/whitespace.c b/src/parsers/whitespace.c
index 04284e86..970a32c8 100644
--- a/src/parsers/whitespace.c
+++ b/src/parsers/whitespace.c
@@ -75,6 +75,7 @@ static const HParserVtable whitespace_vt = {
.isValidCF = ws_isValidCF,
.desugar = desugar_whitespace,
.compile_to_rvm = ws_ctrvm,
+ .higher = false,
};
HParser* h_whitespace(const HParser* p) {
diff --git a/src/parsers/xor.c b/src/parsers/xor.c
index e031d5d5..3a3f21d2 100644
--- a/src/parsers/xor.c
+++ b/src/parsers/xor.c
@@ -36,6 +36,7 @@ static const HParserVtable xor_vt = {
.isValidRegular = h_false,
.isValidCF = h_false, // XXX should this be true if both p1 and p2 are CF?
.compile_to_rvm = h_not_regular,
+ .higher = true,
};
HParser* h_xor(const HParser* p1, const HParser* p2) {
--
GitLab